Confusion over Belichick's non-trade with SF on Pick # 33 in draft

[patchick]

Odds of getting a player of AAW of 34 and above with those picks:

1x #17: 51%
2x #38: 50%
3x #58: 57%
4x #71: 63%
5x #80: 49%
6x #86: 73%
7x #91: 67%
8x #95: 65%
9x #99: 76%
10x #101: 73%

This is awesome. But there's also a major problem: WHEN do you get that player of AAW, and when do you know you've got him? Realistically, the answer isn't always going to be during the rookie's first training camp -- and you only get to keep 53 players.

So it may be far more valuable to take the "4x #71: 63%" who you can actually keep on your roster than the "10x #101: 73%" and have to pick four of the ten out of camp.

 

[plk]

This is awesome. But there's also a major problem: WHEN do you get that player of AAW, and when do you know you've got him? Realistically, the answer isn't always going to be during the rookie's first training camp -- and you only get to keep 53 players.

So it may be far more valuable to take the "4x #71: 63%" who you can actually keep on your roster than the "10x #101: 73%" and have to pick four of the ten out of camp.

These are thought experiments designed to explore whether or not the Draft Value Chart accurately reflects the value of draft picks based on the actual performance of players. They are, of course, impossible trades in real life, and not intended to be taken in that way.

On the previous subject, let me emphasize that it is critical to make trades based on the "ultimate benefit of players contributing to a team." The Draft Value Chart may help you determine what trades you can make in so far as the person that you're trading with believes in it. There is little reason to believe that it is valuing a particular pick, in relation to other picks, accurately. The above calculations suggest that it may not be doing so.

 

[patchick]

These are thought experiments designed to explore whether or not the Draft Value Chart accurately reflects the value of draft picks based on the actual performance of players. They are, of course, impossible trades in real life, and not intended to be taken in that way.

On the previous subject, let me emphasize that it is critical to make trades based on the "ultimate benefit of players contributing to a team." The Draft Value Chart may help you determine what trades you can make in so far as the person that you're trading with believes in it. There is little reason to believe that it is valuing a particular pick, in relation to other picks, accurately. The above calculations suggest that it may not be doing so.

But isn't the impossibility of certain options part of the calculation?

In theory, drafting 30 guys in round 6 gives you a better chance of hitting on one or more than taking 2 guys in round 2 at a much lower "chart price," showing an irrationality in the pricing structure.

In practice, acquiring 30 6th-round picks in a 32-team draft isn't feasible, AND keeping even a fraction of them would cripple your roster. And that presumably fuels the seeming undervaluation of late-round picks.

 

[plk]

But isn't the impossibility of certain options part of the calculation?

In theory, drafting 30 guys in round 6 gives you a better chance of hitting on one or more than taking 2 guys in round 2 at a much lower "chart price," showing an irrationality in the pricing structure.

In practice, acquiring 30 6th-round picks in a 32-team draft isn't feasible, AND keeping even a fraction of them would cripple your roster. And that presumably fuels the seeming undervaluation of late-round picks.

Of course you are right. These calculations are only intended to be suggestive. As well as interesting and amusing.

The only claim is that they suggest that there may be problems with the way that the Draft Value Chart values picks in relation to each other. They don't prove anything, but they sure do suggest that too me.

Drafting is all a matter of probabilities, and the calculations do say something I find interesting about the probabilities involved.

I hope someday we will have better information on the accuracy or inaccuracy of the Draft Value Chart. This is just an attempt that can be made now.

 

[patchick]

I hope someday we will have better information on the accuracy or inaccuracy of the Draft Value Chart.

Agreed, and if I didn't have some semblance of a real life to live I'd love to try to map out real historical outcomes! (And to get back to your earlier point about what goes on in the secret Beli-cave...surely Ernie Adams has done just that.)

 

[plk]

(And to get back to your earlier point about what goes on in the secret Beli-cave...surely Ernie Adams has done just that.)

Absolutely.

 

[MaineMan]

These are thought experiments designed to explore whether or not the Draft Value Chart accurately reflects the value of draft picks based on the actual performance of players. They are, of course, impossible trades in real life, and not intended to be taken in that way.

On the previous subject, let me emphasize that it is critical to make trades based on the "ultimate benefit of players contributing to a team." The Draft Value Chart may help you determine what trades you can make in so far as the person that you're trading with believes in it. There is little reason to believe that it is valuing a particular pick, in relation to other picks, accurately. The above calculations suggest that it may not be doing so.

See, this is where this doesn't work for me.

A draft pick number merely represents a place in line. Teams trade up to a higher pick (closer to the front of the line) not based on some abstract notion that players who may be available at that point will have a higher AV, but for a specific player who the team believes will be selected by a competitor between THAT pick and the next pick that they currently hold. What "price" the team pays in terms of SCV may depend on how highly the team values that specific player at the time, but there are limits that may or may not be related to SVC values (e.g., what other specific prospect targets might the team be losing the opportunity to draft by sacrificing a second or third pick to trade up?).

However, what performance value that player ultimately provides after the pick has been made has exactly zero relationship to the SVC since that value depends entirely on the competence of the drafting team and happenstance. For example, does the #43 pick become worth less than its SVC because Sergio Kindle did a Jack-and-Jill on a flight of stairs?

That makes no sense to me. Do the Ravens get to return Kindle for the 470 SCV points they paid for him "because he was broken"? Of course not. The value he produces once they "own" him is entirely up to them. The 470 they paid was for the opportunity to draft him.

 

[Off The Grid]

I can't wait until I've consumed enough barbiturates to actually follow this discussion.

 

[unoriginal]

This is awesome. But there's also a major problem: WHEN do you get that player of AAW, and when do you know you've got him? Realistically, the answer isn't always going to be during the rookie's first training camp -- and you only get to keep 53 players.

So it may be far more valuable to take the "4x #71: 63%" who you can actually keep on your roster than the "10x #101: 73%" and have to pick four of the ten out of camp.

Reminds me of an old thread on taking QBs in the first round, in which I did some similar analysis.

I really should update that using the handy AAV metric instead of my own subjective biases.

 

[plk]

See, this is where this doesn't work for me.

A draft pick number merely represents a place in line. Teams trade up to a higher pick (closer to the front of the line) not based on some abstract notion that players who may be available at that point will have a higher AV, but for a specific player who the team believes will be selected by a competitor between THAT pick and the next pick that they currently hold. What "price" the team pays in terms of SCV may depend on how highly the team values that specific player at the time, but there are limits that may or may not be related to SVC values (e.g., what other specific prospect targets might the team be losing the opportunity to draft by sacrificing a second or third pick to trade up?).

However, what performance value that player ultimately provides after the pick has been made has exactly zero relationship to the SVC since that value depends entirely on the competence of the drafting team and happenstance. For example, does the #43 pick become worth less than its SVC because Sergio Kindle did a Jack-and-Jill on a flight of stairs?

That makes no sense to me. Do the Ravens get to return Kindle for the 470 SCV points they paid for him "because he was broken"? Of course not. The value he produces once they "own" him is entirely up to them. The 470 they paid was for the opportunity to draft him.

For the record, I agree with your comments, but they are at 90° to the discussion that you have jumped into in the middle of. I know this is easy to do.

The discussion is about the origin of the Draft Value Chart, and why the way that it was constructed may affect its reliability. I made a misguided attempt to try to explain a point made in a chapter on the Draft Value Chart in the book "Scorecasting."

 

[plk]

Reminds me of an old thread on taking QBs in the first round, in which I did some similar analysis.

I am amused to find this exchange in that old thread. It has perhaps has some slight relevance to part of the discussion above.

Based on data in this thread I would have to take a QB with every pick in my draft and hopefully finding one that works with the %s being for the 1st rounder.

Seriously, I'm surprised that some team with a bunch of comp picks hasn't done something along these lines -- pick 3 QBs in rounds 4-7, see if any of them look like keepers by the end of camp.

Can you remember any team drafting even 2 quarterbacks in 1 draft? Why the heck not?

 

[plk]

Here is an attempt here is an attempt to revise the Draft Value Chart based on performance. There seem to be other attempts on the same site. There is a vast amount of accessible data on this site.

The Draft Value Chart: Right or Wrong? » Pro-football-reference.com blog » Blog Archive

 

[patchick]

I am amused to find this exchange in that old thread. It has perhaps has some slight relevance to part of the discussion above.

Wow, I don't even remember that conversation! But let's try it...I'm going to find the last 3 QBs taken in some drafts old enough to judge:

2005: Ryan Fitzpatrick, Matt Cassel, James Kiliam
2006: D.J. Shockley, Bruce Gradkowski, Reggie McNeal
2007: Tyler Thigpen, Jordan Palmer, Troy Smith

Not bad at all! Each year has yielded at least one QB who has started multiple games in the NFL. Now without peeking I'm going to randomly choose...let's say the 1st QB taken in round 3 of each of those years.

2005: Charlie Frye
2006: Charlie Whitehurst
2007: Trent Edwards

Interesting. Realistically, you can't run your whole draft this way -- and the "bulk" approach could look much worse for positions that are easier to project from measurables. But if you're a team seriously in the market for a developmental QB to take over in 1-2 years, piling up on 7th rounders doesn't look like a bad way to go.

 

[plk]

Wow, I don't even remember that conversation! But let's try it...I'm going to find the last 3 QBs taken in some drafts old enough to judge:

2005: Ryan Fitzpatrick, Matt Cassel, James Kiliam
2006: D.J. Shockley, Bruce Gradkowski, Reggie McNeal
2007: Tyler Thigpen, Jordan Palmer, Troy Smith

Not bad at all! Each year has yielded at least one QB who has started multiple games in the NFL. Now without peeking I'm going to randomly choose...let's say the 1st QB taken in round 3 of each of those years.

2005: Charlie Frye
2006: Charlie Whitehurst
2007: Trent Edwards

Interesting. Realistically, you can't run your whole draft this way -- and the "bulk" approach could look much worse for positions that are easier to project from measurables. But if you're a team seriously in the market for a developmental QB to take over in 1-2 years, piling up on 7th rounders doesn't look like a bad way to go.

I was quite amused that there could be a real world application of the "bulk" approach. I had not expected it.

 

[Oswlek]

Here is an attempt here is an attempt to revise the Draft Value Chart based on performance. There seem to be other attempts on the same site. There is a vast amount of accessible data on this site.

The Draft Value Chart: Right or Wrong? » Pro-football-reference.com blog » Blog Archive

As someone who is in finance, I am fully aware of the discrepancy between value and price.

 

[MaineMan]

For the record, I agree with your comments, but they are at 90° to the discussion that you have jumped into in the middle of. I know this is easy to do.

The discussion is about the origin of the Draft Value Chart, and why the way that it was constructed may affect its reliability. I made a misguided attempt to try to explain a point made in a chapter on the Draft Value Chart in the book "Scorecasting."

Okay. Perhaps I am lost.

"Reliability" with regard to what?

 

[patchick]

I was quite amused that there could be a real world application of the "bulk" approach. I had not expected it.

Yep -- though it might be the best and only! FWIW, I did the same little exercise with cornerbacks (as a position with more of a measurables-driven profile), and sure enough it's not pretty:

2005: Chris Roberson, Adrian Ward, Reynaldo Hill
2006: T.J. Rushing, Dee Webb, Justin Phinisee
2007: Marcus Hamilton, Alan Ball, C.J. Wilson

I believe Wilson is the only one of the 9 who has never been cut outright, and even he has never started a single game.

 

[plk]

Yep -- though it might be the best and only! FWIW, I did the same little exercise with cornerbacks (as a position with more of a measurables-driven profile), and sure enough it's not pretty:

2005: Chris Roberson, Adrian Ward, Reynaldo Hill
2006: T.J. Rushing, Dee Webb, Justin Phinisee
2007: Marcus Hamilton, Alan Ball, C.J. Wilson

I believe Wilson is the only one of the 9 who has never been cut outright, and even he has never started a single game.

According to the Draft Value Chart, one mid first round pick is equal to 44 mid sixth round picks, and hitting on a cornerback with a first-round pick is far from guaranteed. I do believe it may be possible to learn some things from contemplating a bulk strategy even though it is impossible to actually use it. Maybe one could derive a usable strategy from reflecting on it.

This chart is by Defensive Backs which is just the way they do it:

All-Time Draft Listing: DB 1980 -- 2010 - Pro-Football-Reference.com

Just eyeballing it, it looks as if there is not much difference between choosing a cornerback in the first round and in the second round. Some years have only a few misses; some have a lot.

Some good Defensive Backs are chosen in last three rounds. I am not up to counting whether it is one in 20, or in 44, or in 108 (which is the mid first round to mid seventh round figure), and it would not prove anything anyway. Hits seem to have been more frequent before 1990, but a number of them come after round seven and would be undrafted free agents now.

 

[plk]

As someone who is in finance, I am fully aware of the discrepancy between value and price.

It really should have been called the Draft Price Chart, but it is too late now.

 

[Off The Grid]

Wow, I don't even remember that conversation! But let's try it...I'm going to find the last 3 QBs taken in some drafts old enough to judge:

2005: Ryan Fitzpatrick, Matt Cassel, James Kiliam
2006: D.J. Shockley, Bruce Gradkowski, Reggie McNeal
2007: Tyler Thigpen, Jordan Palmer, Troy Smith

Not bad at all! Each year has yielded at least one QB who has started multiple games in the NFL. Now without peeking I'm going to randomly choose...let's say the 1st QB taken in round 3 of each of those years.

2005: Charlie Frye
2006: Charlie Whitehurst
2007: Trent Edwards

Interesting. Realistically, you can't run your whole draft this way -- and the "bulk" approach could look much worse for positions that are easier to project from measurables.

But if you're a team seriously in the market for a developmental QB to take over in 1-2 years, piling up on 7th rounders doesn't look like a bad way to go.

GOSH, Sister Pat, why didn't I ever recommend that sort of philosophy??

I seem to remember you coming down pretty emphatically against rolling up Day 3 QB prospects, when I quite recently did so.

Welcome aboard.